tfp_causalimpact_customized 0.1.31

A stable version of the tfcasualimpact package

tfp_causalimpact_customized

Features

• Rebuilt of TFP CausalImpact

Improved summary round to 3 digits

Matplotlib Japanese Support

• Added support for Japanese fonts and characters in Matplotlib plots.
• Enhanced compatibility with Japanese data visualization requirements.

Improved Matplotlib Plots

• Enhanced plotting capabilities for clearer and more informative visualizations.
• Customized plot styles and themes to better represent causal impact analysis.

Comparison with tfcausalimpact

Enhancements Over tfcausalimpact

• Stability: Resolved the issue of results changing from run to run, ensuring consistent outcomes. See Result change from run to run in tfcausalimpact.
• Performance: Optimized performance for faster computations and larger datasets.
• Customization: Increased flexibility in model customization and parameter tuning.

Fixed Issues

• Consistent Results: Fixed the Result change from run to run in tfcausalimpact issue to ensure reproducible results across multiple runs.
• Bug Fixes: Addressed various bugs reported in the original tfcausalimpact repository to enhance overall stability and reliability.

Getting Started

1. Installation

   uv add tfp_causalimpact_customized
   

2. Plot options (Currently only Matplotlib is supported) Important:y_formatter_unit must be a dictionary with the keys that are the same as legend_labels y_labels.

plot_options = {
    'chart_width': 1000,
    'chart_height': 200,
    'x_label': 'Date',
    'y_labels': ['Observed1', 'Pointwise Effect1', 'Cumulative Effect1'],
    'title': 'Customized Matplotlib Plot',
    'title_font_size': 16,
    'axis_title_font_size': 14,
    'y_formatter': 'millions',
    'y_formatter_unit': {
        'Observed1': ' units',
        'Pointwise Effect1': ' effect',
        'Cumulative Effect1': ' total'
    },
    'legend_labels': {
        'mean': 'Average',
        'observed': 'Observed',
        'pointwise': 'Pointwise Effect',
        'cumulative': 'Cumulative Effect',
        'pre-period-start': 'Start of Pre-Period',
        'pre-period-end': 'End of Pre-Period',
        'post-period-start': 'Start of Post-Period',
        'post-period-end': 'End of Post-Period'
    }
}